Lecture 5: Is there a nutrition-based poverty
trap?
Abhijit V. Banerjee and Esther Duflo
14.73 Challenges of World Poverty
Under-nutrition in the World
I
While famine may be history, malnutrition is not.
I
The UN agency FAO estimates that, worldwide a billion
people are under-nourished.
I
Symptoms of malnutrition: anemia, low BMI (bodymass
index), small and thin children.
Large increase in food prices in 2006-2008, and again in 2010.
Two consequences on those of the poor who are net consumer
of food (i.e. they produce less than they consume, e.g. urban
poor).
I
Under-nutrition in the World
I
While famine may be history, malnutrition is not.
I
The UN agency FAO estimates that, worldwide a billion
people are under-nourished.
I
Symptoms of malnutrition: anemia, low BMI (bodymass
index), small and thin children.
Large increase in food prices in 2006-2008, and again in 2010.
Two consequences on those of the poor who are net consumer
of food (i.e. they produce less than they consume, e.g. urban
poor).
I
I
I
Because a larger portion of their budget is spent on food, this
will affects them disproportionately (the real price of their total
food basket has increased more).
Price increase may lead to a decrease in the nutritional status
of the poor and start a vicious circle: Pak Solhin’s story
The S-Shape curve and the nutrition-based poverty trap:
reminder
I
With your wage, you buy food, which gives you strength,
which allows you to get wages: it creates a relationship
between income today, and income tomorrow (or with wage
level, and the ability of the poorest people to work: at the
extremely only those who have some non-labor income and
can supplement their daily wage will be able to work).
I
Necessary condition for a poverty trap: the capacity curve
intersects the 45 degree line from below at some point S-Shape
The S-shape is made of two relations:
I
I
I
The relationship between wage and nutrition (how much better
do you eat if you have a little more income)
And the relationship between nutrition and productivity (how
much stronger to do you become if you have a bit more to eat).
Understanding Food Consumption
I
If there was a S-Shape curve between nutrition and
productivity, the poor should eat as much as they can:
I
I
The share of food in the budget would be very high for them.
If you have some unavoidable expense, expenditure on food
would first increase more than proportionally, and then less
than proportionally.
I
I
I
budget: 20 rupees=5 rupees on clothing and house, 15 rupees
on food
budget: 30 rupees= 5 rupees on clothing and house, 25
rupees on food
budget: 45 rupees=10 rupees on clothing on house, 30 rupees
on food, 10 rupees on movies
Do the poor eat as much as they can?
I
I
The Food share in the budget around the world : Budgets
While the share of food in the budget is fairly high, the poor
have two margins to increase their consumption
Do the poor eat as much as they can?
I
I
The Food share in the budget around the world : Budgets
While the share of food in the budget is fairly high, the poor
have two margins to increase their consumption
I
I
They could spend more on food (see the share spent on other
things, e.g. alcohool, tobacco, etc.)
They could spend the budget they spend on food differently.
Do the poor eat as much as they can?
I
I
The Food share in the budget around the world : Budgets
While the share of food in the budget is fairly high, the poor
have two margins to increase their consumption
I
I
They could spend more on food (see the share spent on other
things, e.g. alcohool, tobacco, etc.)
They could spend the budget they spend on food differently.
I
So are calories increasing very rapidly with income for the very
poor?
I
We do that by getting detailed information from people about
food consumed last month (or last week) and using a calorie
conversion table to estimate how much calories that
represents.
Calorie consumption and economic well being
I
The graph plots the logarithm of calorie consumption against
the logarithm of total household expenses per capital (outlay)
Graphs
I
The slope of this graph is about 0.35.
I
Interpreting this graph: when total expenditure per capita
increase by 1%, the consumption of calories increase by 0.35%
Calorie consumption and economic well being
I
The graph plots the logarithm of calorie consumption against
the logarithm of total household expenses per capital (outlay)
Graphs
I
The slope of this graph is about 0.35.
I
Interpreting this graph: when total expenditure per capita
increase by 1%, the consumption of calories increase by 0.35%
This is an elasticity
Calorie consumption and economic well being
I
The graph plots the logarithm of calorie consumption against
the logarithm of total household expenses per capital (outlay)
Graphs
I
The slope of this graph is about 0.35.
I
Interpreting this graph: when total expenditure per capita
increase by 1%, the consumption of calories increase by 0.35%
This is an elasticity
I
The calories consumed increase with overall consumption:
however, not 1 for 1: when total expenditure increase by 10%,
the consumption of calories increases by about 3.5%. The
Engel Curve
Why is the slope of the Engel curve less than 1?
I
What happens:
Why is the slope of the Engel curve less than 1?
I
What happens:
I
When they get a little more money, people increase the share
of the budget going to other thing: elasticity of overall food
expenditure is 0.7.
Why is the slope of the Engel curve less than 1?
I
What happens:
I
I
When they get a little more money, people increase the share
of the budget going to other thing: elasticity of overall food
expenditure is 0.7.
When they spend more on food, they also buy more expensive
calories (meat instead of cereals; rice instead of coarse cereals:
table the elasticity is price per calorie is also about 0.35.
graph
Why is the slope of the Engel curve less than 1?
I
What happens:
I
I
When they get a little more money, people increase the share
of the budget going to other thing: elasticity of overall food
expenditure is 0.7.
When they spend more on food, they also buy more expensive
calories (meat instead of cereals; rice instead of coarse cereals:
table the elasticity is price per calorie is also about 0.35.
graph
I
In summary: a poor household who is 10% richer spends
about 7% more on food, and this extra spending is shared in
half: half to get more calories, half to get more expensive
(better tasting) calories.
I
Even among very poor people, increase in economic well-being
has a positive, but not huge impact on calories consumed.
Jensen and Miller: In search of a Giffen good
I
What is a Giffen good?
Jensen and Miller: In search of a Giffen good
I
What is a Giffen good? A good whose consumption decreases
when the price decreases.
Jensen and Miller: In search of a Giffen good
I
I
What is a Giffen good? A good whose consumption decreases
when the price decreases.
Two effect when the price of a good decreases
Jensen and Miller: In search of a Giffen good
I
I
What is a Giffen good? A good whose consumption decreases
when the price decreases.
Two effect when the price of a good decreases
I
I
An substitution effect: you want to consume more of this good
because it has become less expensive than other goods.
An income effect
I
I
Normal good: The income effect is positive (you consume
more as income goes up)
Inferior good: The income effect is negative (you consume less
as income goes up)
Jensen and Miller: In search of a Giffen good
I
I
What is a Giffen good? A good whose consumption decreases
when the price decreases.
Two effect when the price of a good decreases
I
I
An substitution effect: you want to consume more of this good
because it has become less expensive than other goods.
An income effect
I
I
I
I
Normal good: The income effect is positive (you consume
more as income goes up)
Inferior good: The income effect is negative (you consume less
as income goes up)
In most cases, even for inferior goods the substitution effect
will dominate, because most goods are only a small part of
the budget.
But for a staple food that constitutes a large part of the
budget, a decrease in the price may have a large income
effect: A giffen good is when the (negative) income effect is
larger than the (positive) substitution effect
How to uncover a Giffen good?
How to uncover a Giffen good?
I
First strategy: observe that, in China, in cities where the price
of rice is higher, people consume more rice.
I
Why did they conclude that this was not proof that rice was a
Giffen good?
How to uncover a Giffen good?
I
First strategy: observe that, in China, in cities where the price
of rice is higher, people consume more rice.
I
Why did they conclude that this was not proof that rice was a
Giffen good?
Second strategy. Conduct a randomized experiment:
I
How to uncover a Giffen good?
I
First strategy: observe that, in China, in cities where the price
of rice is higher, people consume more rice.
I
Why did they conclude that this was not proof that rice was a
Giffen good?
Second strategy. Conduct a randomized experiment:
I
I
I
I
I
Take a sample of households, and randomly chose a subsample
of them.
Distribute vouchers for reduced price of rice in Hunan, reduced
price of wheat in Gangsu to the random subsample, for more
than month supply
Make sure that households do not exchange or trade them
(otherwise it would be a pure income transfer, there would be
no substitution).
After 6 month, ask households detailed questions about their
consumption of rice, wheat, and other stuff.
Main results
I
: when the price of rice decrease by 10%, rice
consumption decrease by 2.5% (elasticity: -0.25).
Hunan
Main results
I
: when the price of rice decrease by 10%, rice
consumption decrease by 2.5% (elasticity: -0.25).
I
In the table, a result that has ** in front of it means that we
can say that it is statistically different from zero. i.e. if we run
100 placebo experiments instead, and each time we take the
difference between treatment and control, and we plot the
distribution of these differences, 97.5% of them will be lower
(or higher) than the difference we find here.
Hunan
Main results
I
: when the price of rice decrease by 10%, rice
consumption decrease by 2.5% (elasticity: -0.25).
I
In the table, a result that has ** in front of it means that we
can say that it is statistically different from zero. i.e. if we run
100 placebo experiments instead, and each time we take the
difference between treatment and control, and we plot the
distribution of these differences, 97.5% of them will be lower
(or higher) than the difference we find here.
I
For which kinds of goods do we see an increase in
consumption?
Hunan
Main results
I
: when the price of rice decrease by 10%, rice
consumption decrease by 2.5% (elasticity: -0.25).
I
In the table, a result that has ** in front of it means that we
can say that it is statistically different from zero. i.e. if we run
100 placebo experiments instead, and each time we take the
difference between treatment and control, and we plot the
distribution of these differences, 97.5% of them will be lower
(or higher) than the difference we find here.
I
For which kinds of goods do we see an increase in
consumption?
I
What have households done?
Hunan
Main results
I
: when the price of rice decrease by 10%, rice
consumption decrease by 2.5% (elasticity: -0.25).
I
In the table, a result that has ** in front of it means that we
can say that it is statistically different from zero. i.e. if we run
100 placebo experiments instead, and each time we take the
difference between treatment and control, and we plot the
distribution of these differences, 97.5% of them will be lower
(or higher) than the difference we find here.
I
For which kinds of goods do we see an increase in
consumption?
I
What have households done?
I
Hunan
Guansu
: Elasticity is positive (not a Giffen good).
Main results
I
: when the price of rice decrease by 10%, rice
consumption decrease by 2.5% (elasticity: -0.25).
I
In the table, a result that has ** in front of it means that we
can say that it is statistically different from zero. i.e. if we run
100 placebo experiments instead, and each time we take the
difference between treatment and control, and we plot the
distribution of these differences, 97.5% of them will be lower
(or higher) than the difference we find here.
I
For which kinds of goods do we see an increase in
consumption?
I
What have households done?
I
I
Hunan
Guansu
: Elasticity is positive (not a Giffen good).
What explanation do they give for the different results?
Implications for nutrition
I
Many countries use food price subsidies to encourage greater
nutrition. For example India recently introduced nationally a
subsidy scheme for rice in rice consuming regions.
Implications for nutrition
I
Many countries use food price subsidies to encourage greater
nutrition. For example India recently introduced nationally a
subsidy scheme for rice in rice consuming regions.
I
If households consume less rice and more shrimps, but shrimps
are not very nutritious per dollar spent, the effect on calorie
consumption of subsidizing rice may not be large, and it may
even be negative.
Implications for nutrition
I
Many countries use food price subsidies to encourage greater
nutrition. For example India recently introduced nationally a
subsidy scheme for rice in rice consuming regions.
I
If households consume less rice and more shrimps, but shrimps
are not very nutritious per dollar spent, the effect on calorie
consumption of subsidizing rice may not be large, and it may
even be negative.
I
This is what they find in Hunan
table.
Implications for nutrition
I
Many countries use food price subsidies to encourage greater
nutrition. For example India recently introduced nationally a
subsidy scheme for rice in rice consuming regions.
I
If households consume less rice and more shrimps, but shrimps
are not very nutritious per dollar spent, the effect on calorie
consumption of subsidizing rice may not be large, and it may
even be negative.
I
This is what they find in Hunan
I
And this is true not only for calories but also other nutrients.
table.
Implications for nutrition
I
Many countries use food price subsidies to encourage greater
nutrition. For example India recently introduced nationally a
subsidy scheme for rice in rice consuming regions.
I
If households consume less rice and more shrimps, but shrimps
are not very nutritious per dollar spent, the effect on calorie
consumption of subsidizing rice may not be large, and it may
even be negative.
I
This is what they find in Hunan
I
And this is true not only for calories but also other nutrients.
I
What does this tell us about the effect income on calorie
consumption in this population?
table.
Implications for nutrition
I
Many countries use food price subsidies to encourage greater
nutrition. For example India recently introduced nationally a
subsidy scheme for rice in rice consuming regions.
I
If households consume less rice and more shrimps, but shrimps
are not very nutritious per dollar spent, the effect on calorie
consumption of subsidizing rice may not be large, and it may
even be negative.
I
This is what they find in Hunan
I
And this is true not only for calories but also other nutrients.
I
What does this tell us about the effect income on calorie
consumption in this population?
I
It must be negative.
table.
Are the calorie Engel Curve overestimated?
I
How can we reconcile this result with the Engel curve we saw
in India?
Are the calorie Engel Curve overestimated?
I
I
How can we reconcile this result with the Engel curve we saw
in India?
Note that we are comparing different households, not the
same households as we give them more money. This may lead
us to over-estimate the impact of well-being on calorie
consumption
Are the calorie Engel Curve overestimated?
I
I
How can we reconcile this result with the Engel curve we saw
in India?
Note that we are comparing different households, not the
same households as we give them more money. This may lead
us to over-estimate the impact of well-being on calorie
consumption
I
I
Those who eat more may be more productive and have more
money (reverse causality)
Those who have more income may be have different tastes,
and would may be eat more even if they were poorer (for
example, people who smoke may both eat less and earn less).
The Engel Curve over time
I
When we plot the Engel Curve over time in India (different
years), we see that they fall down. graph
The Engel Curve over time
I
When we plot the Engel Curve over time in India (different
years), we see that they fall down. graph
I
With economic growth, household move up the Engel curve,
but the Engel Curve falls: on balance the consumption of
calories and all other nutrients has fallen in India over the last
25 years.
The Engel Curve over time
I
When we plot the Engel Curve over time in India (different
years), we see that they fall down. graph
I
With economic growth, household move up the Engel curve,
but the Engel Curve falls: on balance the consumption of
calories and all other nutrients has fallen in India over the last
25 years. table
The Engel Curve over time
I
When we plot the Engel Curve over time in India (different
years), we see that they fall down. graph
I
With economic growth, household move up the Engel curve,
but the Engel Curve falls: on balance the consumption of
calories and all other nutrients has fallen in India over the last
25 years. table
I
What could be happening?
Does eating more make people more productive?
I
Why are poor households not eating more and not seizing
every available opportunity to eat more?
Does eating more make people more productive?
I
Why are poor households not eating more and not seizing
every available opportunity to eat more?
I
First hypothesis: may be they don’t need to eat as much
Does eating more make people more productive?
I
Why are poor households not eating more and not seizing
every available opportunity to eat more?
I
First hypothesis: may be they don’t need to eat as much
I
Over time, in India, need for calories have gone down: less
“backbreaking work”, fewer illnesses.
I
Households in China where poor urban households.
Does eating more make people more productive?
I
A study by John Strauss: impact of calories on productivity in
Sierra Leone (results do not come an experiment, but Strauss
used the fact that people eat less when the price of food goes
up).
I
Calories makes people more productive, but it looks
like an inverted L-shape curve: it is highest for the poorest:
when their calorie consumption increase by 1%, their
productivity increase by 0.4%, and after that it goes down.
Result
Conclusion
I
I
I
I
I
I
Poor people do not behave as if there were a nutrition based
poverty trap
They do not consume as much calories as they could, and
when their income goes up, they don’t seem to eat that much
more.
Indeed, it seems that the benefit to consuming more calories
may be positive, but perhaps not large to generate a S-Shape,
especially for people who work for a wage.
In Summary, at the maximum: when your income today
increase by 10%, you calorie consumption increases by 3.5%,
and that leads to an increase in productivity (and hence
income tomorrow) of 3.5*0.4=1.4%: positive, but not steep
enough for a poverty trap!
However, other micronutrients may generate a poverty trap
Things may also be very different for children
!"#$%&'$("")$*('+,$%&'-)$."+'/
What do the poor spend their money on?
As a Share of Total Consumption
Alcohol/
Food
Tobacco
Education
Living on less than $1 a day
Rural
Cote d'Ivoire
Guatemala
India - Udaipur
Indonesia
Mexico
Nicaragua
Pakistan
Panama
Papua New Guinea
Peru
South Africa
64.4%
65.9%
56.0%
66.1%
49.6%
57.3%
67.3%
67.8%
78.2%
71.8%
71.5%
2.7%
0.4%
5.0%
6.0%
8.1%
0.1%
3.1%
4.1%
1.0%
2.5%
5.8%
0.1%
1.6%
6.3%
6.9%
2.3%
3.4%
2.5%
1.8%
1.9%
0.8%
Health
2.2%
0.3%
5.1%
1.3%
0.0%
4.1%
3.4%
4.0%
0.3%
0.4%
0.0%
Calorie and Total expenditure per capita
8.28.0-
C.6CU *
7.6-
slid
/
line is calorie outlay regression
7.40
to
brokenlines are two standarderror
bandswith no allowance for cluster design;
outer bandsallow for cluster design.
sinner
4o
>a%<
7.-
/
7.0
i
3.4
3.8
4.2
4.6
5.0
5.4
5.8
6.2
log of percapitaoutlay
FIG. 2.-Regression
rashtra, India, 1983.
1.0
function for log calories and log per capita expenditure, Maha-
two bootstrappedstandarderrors
allowing for clustereffects
Calorie and Total expenditure per capita
8.28.0-
C.6CU *
7.6-
slid
/
line is calorie outlay regression
7.40
to
brokenlines are two standarderror
bandswith no allowance for cluster design;
outer bandsallow for cluster design.
sinner
4o
>a%<
7.-
/
7.0
i
3.4
3.8
4.2
4.6
5.0
5.4
5.8
6.2
log of percapitaoutlay
FIG. 2.-Regression
rashtra, India, 1983.
1.0
function for log calories and log per capita expenditure, Maha-
two bootstrappedstandarderrors
allowing for clustereffects
Deaton and Subramanian, Table 1
How the food budget changes with wellbeing
0.5-
~ ~ ~ ~ ~ ~ ~~~~~
ohs-~
0.4
0.3
'v
t
line+1oiregression
\re
C0.0
~0.2
-0.1-
o
-
--
~
C.)
0
s
-
- -
'X
errors
line+I- twostandard
~~~~~~~~~regression
-
0.0,
-0.1
-
-0.2
3.4
3.8
4.2
4.6
5.0
5.4
5.8
6.2
log of percapitaoutlay
FIG. 4.-Log
India, 1983.
of price per calorie and log of per capita expenditure, Maharashtra,
0.6errors
elasticityplustwostandard
C
0.5 -
Elasticity of consumption of various items with respect to
price subsidy: Hunan
Table 4. Consumption Response to the Price Subsidy
HUNAN
%Subsidy(rice)
%ΔEarned
%ΔUnearned
%ΔPeople
Constant
Observations
R2
%Subsidy(wheat)
%ΔEarned
%ΔUnearned
Rice
Other Cereal
Fruit & Veg
Meat
Seafood
Pulses
Food Out
Non-Food
0.397
(0.355)
-0.001
(0.040)
-0.087
(0.065)
0.46**
(0.19)
7.5***
(2.5)
-0.623***
(0.227)
0.058***
(0.021)
-0.018
(0.040)
0.63***
(0.11)
-0.3
(1.4)
0.377
(0.415)
0.002
(0.043)
0.076
(0.071)
0.05
(0.24)
-5.7**
(2.8)
0.482**
(0.230)
0.036
(0.022)
-0.004
(0.042)
-0.07
(0.10)
-0.2
(1.4)
-0.791*
(0.476)
-0.052
(0.050)
-0.037
(0.075)
0.48**
(0.23)
8.8***
(3.0)
Dairy
-0.054
(0.069)
-0.006
(0.004)
-0.021
(0.019)
0.09
(0.05)
0.2
(0.6)
Fats
-0.235*
(0.140)
0.043***
(0.014)
-0.044*
(0.025)
0.89***
(0.08)
4.1***
(1.0)
-0.567*
(0.313)
0.022
(0.031)
-0.007
(0.055)
0.88***
(0.16)
-8.3***
(2.1)
0.117
(0.347)
0.059
(0.044)
0.020
(0.057)
-0.18
(0.18)
-3.5
(2.5)
0.200
(0.200)
0.014
(0.025)
0.089**
(0.038)
0.15
(0.13)
-52.6***
(1.5)
1258
0.19
1258
0.06
1258
0.11
1258
0.07
1258
0.02
1258
0.03
1258
0.02
1258
0.09
1258
0.02
1258
0.20
Wheat
Other Cereal
Fruit & Veg
Meat
GANSU
Seafood
Pulses
Food Out
Non-Food
-0.283
(0.335)
-0.067
(0.049)
0.130
(0.106)
0.049
(0.190)
0.061**
(0.027)
0.046
(0.077)
0.130
(0.299)
0.085*
(0.044)
0.314***
(0.091)
-0.017
(0.017)
0.000
(0.000)
0.025
(0.025)
0.240
(0.320)
-0.047
(0.043)
0.012
(0.104)
Dairy
0.282
(0.207)
-0.025
(0.029)
0.108
(0.073)
Fats
0.353
(0.258)
0.079**
(0.036)
-0.017
(0.092)
0.507**
(0.251)
0.091***
(0.033)
-0.110
(0.091)
0.109
(0.276)
0.070
(0.043)
-0.077
(0.097)
-0.021
(0.180)
0.040
(0.025)
0.229***
(0.070)
***
*
***
**
Table 4. Consumption Response to the Price Subsidy
HUNAN
Rice
%Subsidy(rice)
Other Cereal
Fruit & Veg
Meat
Seafood
Pulses
Food Out
Non-Food
0.397
(0.355)
-0.001
(0.040)
-0.087
(0.065)
0.46**
(0.19)
7.5***
(2.5)
-0.623***
(0.227)
0.058***
(0.021)
-0.018
(0.040)
0.63***
(0.11)
-0.3
(1.4)
0.377
(0.415)
0.002
(0.043)
0.076
(0.071)
0.05
(0.24)
-5.7**
(2.8)
0.482**
(0.230)
0.036
(0.022)
-0.004
(0.042)
-0.07
(0.10)
-0.2
(1.4)
-0.791*
(0.476)
-0.052
(0.050)
-0.037
(0.075)
0.48**
(0.23)
8.8***
(3.0)
Dairy
-0.054
(0.069)
-0.006
(0.004)
-0.021
(0.019)
0.09
(0.05)
0.2
(0.6)
Fats
-0.235
(0.140)
0.043***
(0.014)
-0.044*
(0.025)
0.89***
(0.08)
4.1***
(1.0)
-0.567*
(0.313)
0.022
(0.031)
-0.007
(0.055)
0.88***
(0.16)
-8.3***
(2.1)
0.117
(0.347)
0.059
(0.044)
0.020
(0.057)
-0.18
(0.18)
-3.5
(2.5)
0.200
(0.200)
0.014
(0.025)
0.089**
(0.038)
0.15
(0.13)
-52.6***
(1.5)
1258
0.19
1258
0.06
1258
0.11
1258
0.07
1258
0.02
1258
0.03
1258
0.02
1258
0.09
1258
0.02
1258
0.20
Wheat
Other Cereal
Fruit & Veg
Meat
GANSU
Seafood
Pulses
Food Out
Non-Food
-0.283
(0.335)
-0.067
(0.049)
0.130
(0.106)
0.52*
(0.29)
23.8***
(2.8)
0.049
(0.190)
0.061**
(0.027)
0.046
(0.077)
1.01***
(0.15)
11.0***
(1.6)
0.130
(0.299)
0.085*
(0.044)
0.314***
(0.091)
-0.10
(0.28)
2.4
(2.5)
-0.017
(0.017)
0.000
(0.000)
0.025
(0.025)
-0.01
(0.01)
-0.2
(0.2)
0.240
(0.320)
-0.047
(0.043)
0.012
(0.104)
0.44**
(0.18)
6.0**
(2.6)
Dairy
0.282
(0.207)
-0.025
(0.029)
0.108
(0.073)
0.10
(0.12)
-3.4*
(1.9)
Fats
0.353
(0.258)
0.079**
(0.036)
-0.017
(0.092)
0.58***
(0.22)
-26.1***
(2.3)
0.507**
(0.251)
0.091***
(0.033)
-0.110
(0.091)
0.66
(0.15)
7.2
(2.1)
0.109
(0.276)
0.070
(0.043)
-0.077
(0.097)
0.00
(0.19)
7.5***
(2.4)
-0.021
(0.180)
0.040
(0.025)
0.229***
(0.070)
-0.04
(0.19)
-38.2***
(1.4)
*
Elasticity of consumption of various items with respect to
price subsidy: Gansu
%ΔEarned
%ΔUnearned
%ΔPeople
Constant
Observations
R2
%Subsidy(wheat)
%ΔEarned
%ΔUnearned
%ΔPeople
Constant
Observations
1269
1269
1269
1269
1269
1269
1269
1269
1269
1269
R2
0.08
0.06
0.07
0.05
0.03
0.06
0.03
0.07
0.05
0.17
Notes: Regressions include county*time fixed-effects. The dependent variables are the arc percent change in household consumption of the good listed at the top of t
column. Standard errors clustered at the household level. %Subsidy(rice/wheat) is the rice or wheat price subsidy, measured as a percentage of the average pri
%ΔEarned is the arc percent change in the household earnings from work; %ΔHH Unearned is the arc percent change in the household income from unearned sourc
(government payments, pensions, remittances, rent and interest from assets); %ΔPeople is the arc percent change in the number of people living in the househo
*Significant at 10 percent level. **Significant at 5 percent level. ***Significant at 1 percent level.
Calories and rice price subsidy
Table 2. Calorie Response to the Price Subsidy
GANSU
HUNAN
%Subsidy(rice/wheat)
%ΔEarned
%ΔUnearned
%ΔPeople
Constant
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
Full
Sample
(Calories)
Below
Median
(Calories)
Above
Median
(Calories)
Bottom
Quartile
(Calories)
Full
Sample
(Protein)
Full
Sample
(Calories)
Below
Median
(Calories)
Above
Median
(Calories)
Bottom
Quartile
(Calories)
Full
Sample
(Protein)
0.004
(0.207)
0.037*
(0.021)
-0.037
(0.034)
1.04***
(0.10)
2.8*
(1.5)
-0.096
(0.133)
0.040***
(0.013)
-0.010
(0.023)
0.93***
(0.07)
0.8
(0.9)
0.154
(0.100)
0.028**
(0.014)
0.046
(0.035)
0.91***
(0.08)
-1.9
(0.8)
0.169
(0.143)
0.027
(0.021)
0.020
(0.056)
1.01***
(0.10)
0.1
(1.1)
0.132
(0.138)
0.029
(0.019)
0.071*
(0.043)
0.81***
(0.12)
-3.9
(1.1)
0.070
(0.261)
0.053
(0.034)
0.101
(0.119)
1.08***
(0.13)
0.6
(1.7)
0.091
(0.112)
0.017
(0.016)
0.069
(0.033)
0.88***
(0.09)
-4.0
(0.9)
-0.206*
(0.108)
0.031***
(0.011)
-0.022
(0.020)
0.94***
(0.07)
0.9
(0.8)
-0.042
(0.144)
0.026*
(0.014)
-0.025
(0.027)
1.07***
(0.08)
1.6
(1.1)
-0.339**
(0.164)
0.036**
(0.018)
-0.023
(0.028)
0.80
(0.11)
0.5***
(1.1)
Observations
1258
633
625
317
1258
1269
634
635
320
1269
R2
0.26
0.34
0.21
0.39
0.20
0.18
0.23
0.15
0.29
0.16
Notes: Regressions include county*time fixed-effects. The dependent variable in columns 1-4 and 6-9 is the arc percent change in household caloric intake
and in columns 5 and 10 it is the arc percent change in household protein consumption. Standard errors clustered at the household level. %Subsidy
(rice/wheat) is the rice or wheat price subsidy, measured as a percentage of the average price. %ΔEarned is the arc percent change in the household earnings
from work; %ΔHH Unearned is the arc percent change in the household income from unearned sources (government payments, pensions, remittances, rent
and interest from assets); %ΔPeople is the arc percent change in the number of people living in the household. *Significant at 10 percent level.
**Significant at 5 percent level. ***Significant at 1 percent level.
Deaton and Dreze, Figure 1
Calorie Engel curves, Rural and Urban India, 1983 to 2004-05
Log per capita household calories
8
1983 rural
1987-8 rural
1993-4 rural
1999-00 rural
7.8
2004-05 rural
7.6
2004-05 urban
1999-00 urban
1993-4 urban
1987-8 urban
1983 urban
7.4
7.2
4.5
5
5.5
6
6.5
7
Logarithm of household per capita expenditure.
Figure by MIT OpenCourseWare.
eaton and Dreze, Table 5
Fractions of the Population Living in Households
with per Capita Calorie Consumption below 2,100
Urban and 2,400 Rural
Year
Round
Rural
Urban
All India
1983
38
66.1
60.5
64.8
1987-8
43
65.9
57.1
63.9
1993-4
50
71.1
58.1
67.8
1999-0
55
74.2
58.2
70.1
2004-5
61
79.8
63.9
75.8
Source: Authors' calculations based on NSS data.
Figure by MIT OpenCourseWare.
John Strauss: Nutrition and Farm Productivity